Stable Diffusion for Furniture Placement with Preserved Room Layout in Home Staging

Sho Hattori, Youiti Kado, Koji Matsumura, Toshihiko Yamasaki

Abstract: This paper introduces a novel approach that combines various technologies including LoCon, ControlNet, and Mobile Line Segment Detection (MLSD) with Stable Diffusion to generate images featuring furniture placements while preserving the integrity of room layouts. The visual presentation of room on property websites significantly impacts users’ room selection choices. The art of strategically arranging furniture, plants, and other items within a room to enhance its appeal is known as home staging. Beyond simply observing the room layout, users are more likely to show interest in renting a room that has undergone effective home staging. While the conventional method of enhancing room images involves image editing, it often fails to preserve room layouts. Our experimental results demonstrate that each of the three image generation patterns (Text2Image, Image2Image, and Inpainting) was capable of generating images that retain the room layout and feature furniture placements. However, achieving the desired output within a single iteration remains a challenge.